Research On The 3D Metric Reconstruction Based On Geometry Scene In Crime Scene Investigation

3D reconstruction belongs to the intermediate vision part in the framework of Marr computing theory and the primary researching content in computer vision. Research on 3D reconstruction and visualization technology has important academic meanings and application values in crime scene investigation, virtual reality and etc. In the thesis, 3D metric reconstruction is deeply analyzed and researched in its theory and application. The main researching contents as follow:First, the camera calibration techniques are analyzed and researched. The camera calibration methods play a critical role in the technology of 3D reconstruction. On the basis of analyzing camera models and imaging principles, various camera calibration methods are analyzed and compared. And usage scopes and application conditions of the calibration methods are provided.Second, a novel camera self-calibration method is proposed, under the condition that the camera's distorted skews and principal points are already known, and the other intrinsic parameters keep varying. The camera must recalibrate after the camera's position's adjustment. So the above method is proposed in the thesis. That is, to calculate the fundamental matrix between images to obtain projective reconstruction at first, and on which basis then, to regain the homography matrix by linear method, at last, to obtain the camera's intrinsic parameters using homography matrix. Experiment results verify the validity of the method.Third, a 3D metric reconstruction method with single view is proposed. The 3D metric reconstruction of scene image is often aimed to the image sequences, and the stratified metric reconstruction method is to obtain the projective reconstruction of image sequences at first. A 3D metric reconstruction method of single view which does not need to obtain the projective reconstruction is proposed. First, the 3D metric reconstructions of different plans are acquired based on the 3D metric reconstructions of planar scene images. Then the length ratios and relative orientations of plans are computed. The 3D metric reconstruction of 3D objects is gained. Experiment results show the validity of the method.The research work is funded by the Ministry of Public Security B level key project 'the Visual Platform of Crime Scene Investigation' (NO.20036125301).